In response to demands that he produce the evidence for his claim that immigration is driving up house prices, the housing minister has now published the calculations. Do they back up his argument?
First, it must be said that researching the drivers of house price increases is fraught with difficulty as there are so many factors involved – not least a degree of sometimes irrational human behaviour given that houses are not merely places to live in but for many are an investment, a source of income or bought in order to speculate on future prices. Especially in the past few years, would-be owner-occupiers have been operating in the same market as a rapidly expanding private rented sector. So if projecting the demand for housing in terms of the number of units is a challenge, translating that into effects on house prices is much more so.
What strikes the reader immediately about the new calculation is its simplicity. The original piece of work produced what was really a rule-of-thumb relationship between household growth and house prices: if the former increases by 1%, the latter will go up by 2%. The context – the years up to 2008 when the research was published – was a period of very strong house price growth.
The method gave satisfactory results for this period but would have faltered immediately afterwards, as prices fell when the global financial crisis occurred, yet population and household numbers continued to grow. What the rule of thumb didn’t attempt to do was explain the causes of house price rises.
So the original model was of its time and based on what were then recent trends. What about the updated version? This has an immediate weakness compared with the earlier one – it assumes household numbers will rise in step with population growth. While this might superficially seem logical, in practice it simply doesn’t hold true: there may be more people in the UK (whether from natural change or net migration), but whether they set up separate households depends on a range of factors, not least the economy. Instead of forming new households, people may share, young people may delay getting married or leaving the parental home, and more may live in overcrowded conditions.
There is another key assumption in the updated version that is highly questionable: it assumes that new migrants form households at the same rate as the indigenous population. But we know that new migrants are more likely to share, that there are many more young people who won’t (for example) be able to afford their own flat, etc. So to assume migrants have the same impact on household growth as an increase in the indigenous population is almost certainly an overstatement, as a footnote to the text obliquely acknowledges.
With the updated version of the model, taking together natural and migrant-led population growth, the growth in real incomes and increase in housing supply, all over 25 years, the increase in real house prices (after inflation) of about £70,000 since 1991 appears to be satisfactorily ‘explained’.
But there is a huge gap in this calculation: the economics of buying a house have changed enormously in the last decade. Interest rates are much lower, but the availability of 100% mortgages is also much lower; the size of typical deposits is much larger and purchase costs have changed. Government intervention in the market is much greater – not least through Help to Buy. Have none of these factors influenced house prices? No one would believe that they haven’t, yet the model ignores their direct effects.
We return to the point we made at the beginning. When this model was drawn up, house prices had been growing rapidly at a time when the factors influencing them had more or less changed in step with each other. But since 2008 the world has altered radically, and the housing market of 2018 is very different.
Using a model that worked a decade ago to attribute a large proportion of house price inflation to immigration, without explicitly taking account of the new drivers that now apply, is foolhardy.
Original post and comments: Inside Housing